Make with Notion 2025: Building an AI-first business (Jeanne DeWitt Grosser)

TL;DR

AI-first advantage comes from redesigning roles, workflows, and value creation—not from adding AI features to existing processes.

Briefing Cornell Notes

Briefing

AI-first businesses are winning by treating artificial intelligence as an operating model—not a bolt-on feature—and that shift forces companies to redesign roles, workflows, and product experiences end to end. Jeanne DeWitt Grosser, COO of Verscell, frames the core lesson as a convergence of “docs” and “demos”: written systems and visual execution should feed each other so teams can move fast without losing alignment or quality. In her view, the companies that pull ahead aren’t merely adopting new tools; they’re reimagining how work gets done, how value gets created, and how institutional knowledge becomes usable at scale.

At Verscell, that reimagining is tied to measurable outcomes and a platform approach. The company has introduced an “AI cloud” described as a unified platform for building, deploying, and running intelligent applications and agents. Verscell reports 80% year-on-year growth and crossing 200 million in ARR, with “v0” still in early stages but already drawing more than half its customers from large enterprises. Speed remains a signature, but the challenge becomes maintaining it across many product stakeholders while ensuring quality and alignment. Notion is presented as the connective tissue: product area plans link high-level strategy to actual work, while Notion AI distills long documents into executive summaries and converts technical documentation into customer-ready messaging drafts for go-to-market teams. Verscell attributes operational gains to this system, including 35% faster shipping, 89% confidence in quality, and nine hours saved per person weekly.

The people side of the AI-first shift starts with hiring roles that didn’t exist a year earlier. One example is “go-to-market engineering,” where go-to-market is treated like a product. The function focuses on exploring, validating, and incubating technical ideas into scalable playbooks; shipping revenue-driving code using AI and data; and redefining go-to-market by “dogfooding” Verscell products and showing customers how agents can reshape their businesses. A concrete case: an inbound SDR workflow was rebuilt into an agent-based play by pairing a top-performing SDR with a go-to-market engineer. The effort took six weeks, reduced touches to reach an opportunity from eight to four, kept opportunity conversion steady, and—after back-testing—claimed 99.5% accuracy when disqualifying inbound leads.

Practices are where “context engineering” becomes central. Verscell runs company-wide demo days every Friday, with every function—from go-to-market to legal, HR, and finance—using Notion pages where an agent summarizes outcomes and next steps and preserves artifacts embedded from v0. That documentation is not just recordkeeping; it becomes training data for agents. For go-to-market specifically, SDRs draft emails in Notion after researching target companies, then a Slackbot route sends drafts to be edited to match executive voice and tone. Over 30 days, the system builds a corpus of “gene-approved” emails, which the company uses to coach SDRs and gradually increase confidence so agents can send more autonomously.

On products, the argument turns to software architecture. Fixed-schema software that forces users into a vendor’s workflow is portrayed as increasingly obsolete. Composable systems and autonomous agents should enable per-user, per-company workflows—so Notion can have its own “Salesforce-like” front end, tailored to its business, while still relying on shared systems of record. The broader trajectory runs from assistants and chatbots to agents that automate processes, and eventually to autonomous systems that orchestrate end-to-end workflows with minimal oversight. The final warning is strategic: AI can’t be layered onto a flawed process. Building AI during “an earthquake” means the ground is moving—so companies must make AI the operating model, not the feature layer.

Cornell Notes

AI-first companies win by turning AI into an operating model that reshapes roles, workflows, and product design. Verscell credits Notion for standardizing alignment (roadmaps and plans) and for converting messy inputs and long documentation into structured, customer-ready outputs, reporting gains like faster shipping and time saved. The company also builds new roles such as “go-to-market engineering,” treating go-to-market like a product and using agent-based playbooks to improve qualification and conversion. A key practice is “context engineering,” where teams capture not just what they do but why it’s done—using Notion pages, demo days, and agent summaries to create a reusable knowledge base. The product direction points toward composable systems and autonomous agents that can orchestrate workflows end to end.

Why does the talk treat AI as more than a feature?

AI is framed as an operating model because it changes how companies run work: it rearranges roles, workflow design, and how value is created. Verscell’s examples connect this to concrete changes—new hiring categories (like go-to-market engineering), standardized planning systems, and agent-driven documentation that feeds execution. The message is that adopting AI tools without changing the underlying way teams operate won’t deliver durable advantage.

How does Verscell’s “go-to-market engineering” work, and what problem does it solve?

Go-to-market engineering treats go-to-market as a product. The function explores, validates, and incubates technical ideas into scalable playbooks; ships revenue-driving code using AI and data; and redefines go-to-market by dogfooding Verscell products and showing customers how agents can reshape their sales motions. A specific result: an inbound SDR workflow was converted into an agent-based play by pairing a top SDR with a go-to-market engineer, taking six weeks, cutting touches to reach an opportunity from eight to four while keeping conversion steady, and claiming 99.5% accuracy when disqualifying inbound leads after back-testing.

What is “context engineering,” and how is it applied to SDR outreach?

Context engineering means systematically capturing not only actions but the reasoning and standards behind them—what excellent looks like. In the SDR example, SDRs research target companies, draft emails in Notion, and then route drafts through a Slackbot for editing by the executive voice (the talk describes the leader revising tone and “altitude”). After 30 days, the company builds a corpus of “gene-approved” emails for different target types (e.g., e-commerce and SaaS), then uses it to coach SDRs so they learn to write like the executive. The longer-term goal is to increase confidence so agents can send more autonomously by Q4.

How do Notion and demo days reinforce each other in Verscell’s operating system?

Verscell holds hour-long demo days every Friday, with 12 demo sessions weekly and participation from every function, not just engineering. Each demo has a Notion page where an agent summarizes outcomes and next steps, and v0 embeds and preserves artifacts. That creates “contextual documentation” that serves as company memory and also supplies the structured inputs agents need to learn and improve their job performance.

What does the talk predict about software interfaces and why does it matter?

The talk argues that fixed-schema software that forces users into a vendor’s workflow is “dead,” because composable software and autonomous agents can deliver per-company and per-user workflows. It uses the analogy of headless commerce: different retail brands can have different front ends while sharing the same underlying systems of record. The implication is that customization will spread beyond retail into other verticals, and AI agents will increasingly automate or orchestrate workflows across systems.

What strategic warning is given for building an AI-first business?

AI can’t be slapped onto a bad process. The talk compares it to building a house during an earthquake: the ground is always moving, so companies must make AI their operating model and understand best practices in their ways of working before layering AI on top.

Review Questions

Which specific operational changes (roles, workflows, documentation practices) does Verscell use to support the claim that AI is an operating model?
In the inbound SDR case, what metrics improved, and what method was used to build the agent-based playbook?
How does “context engineering” differ from simply collecting data, and what mechanisms (Notion pages, demo days, Slackbot editing) make it work in practice?

Key Points

1
AI-first advantage comes from redesigning roles, workflows, and value creation—not from adding AI features to existing processes.
2
Verscell reports measurable gains tied to Notion: 35% faster shipping, 89% confidence in quality, and nine hours saved per person weekly.
3
“Go-to-market engineering” treats sales execution like a product, using AI and data to build and iterate agent-based playbooks.
4
Pairing a top SDR with a go-to-market engineer enabled an inbound workflow rebuild that reduced touches to opportunities from eight to four while keeping conversion steady.
5
Context engineering captures the reasoning and standards behind excellent work, using Notion drafts, agent summaries, and human-in-the-loop edits to train better outputs.
6
Demo days function as a company-wide feedback loop, with Notion pages and v0 artifacts turning execution into structured, reusable knowledge.
7
Composable systems and autonomous agents are positioned to replace fixed-schema software that forces users into a vendor’s workflow.

Highlights

AI is framed as an operating model that restructures how companies work, not a standalone product feature.

Verscell’s inbound agent playbook cut SDR touches to reach an opportunity from eight to four while maintaining opportunity conversion.

Context engineering for SDR outreach uses Notion drafting plus Slackbot routing to preserve executive voice and build a reusable email corpus.

Demo days across every function feed Notion pages where agents summarize outcomes and preserve v0 artifacts, creating “contextual documentation.”

The future direction points from assistants to agents and toward autonomous systems that orchestrate end-to-end workflows with minimal oversight.

Topics

AI-First Operating Model
Go-To-Market Engineering
Context Engineering
Agent-Based Workflows
Composable Software

Mentioned

Notion
Verscell
V0
AI Cloud
Slackbot
Jeanne DeWitt Grosser
ARR
GTM
SDR
CTO
CPO
AI